Customer Service Agent · Use case
How can AI draft Zendesk or ServiceNow replies without sending customer data to a public AI service?
A customer service agent reads a ticket in Zendesk, ServiceNow or Salesforce, finds the help articles and past cases that bear on it, and drafts a reply with its sources. It drafts and a person sends: the agent does not reply to a customer or update a ticket. It runs on your own infrastructure, so customer data is not sent to a public AI service.
The same answers, typed again.
And data that cannot leave.
- ✗ 01Support staff answer the same questions many times, searching help articles and old tickets for each reply.
- ✗ 02Tickets hold customer data, so a hosted assistant is ruled out, and a reply sent with no person reading it is a risk few teams accept.
Where it stops for a person
Before any reply is sent.Before any reply is sent. Every draft is read, edited and sent by a member of the support team.
It drafts.
A person sends.
Step · 01
The ticket system and the help articles are connected.
Step · 02
For each ticket, the agent reads the request and its history.
Step · 03
It retrieves the help articles and past cases that bear on the request.
Step · 04
It drafts a reply and lists the sources it used.
Step · 05
A member of the support team edits the draft and sends it from the ticket system.
Reads from · connectors
The agent connects to Zendesk, ServiceNow or Salesforce through connectors in the Katonic platform. Freshdesk, Intercom and Help Scout connect in the same way.
Acts through · MCP servers
Nothing: it drafts and a person sends. The agent does not write to the ticket system.
What you set up
- Connect the ticket system and the help articles.
- Set the tone, and the subjects the agent should pass to a person without drafting.
- Score drafts against past tickets before release.
Where it runs
The agent, its knowledge store and the model run on infrastructure you own or control. You choose the model and serve it yourself, so questions, documents and answers are not sent to a public AI service.
How you get it today
This page is labelled Use case: the agent is built in Katonic Studio. You describe the job, connect what is named above, add test questions and release the agent once it passes them. When the Catalog holds a tested template for this job, this page will carry the Template label and show how to take it.
Does the agent reply to customers by itself?
No. It drafts a reply, and a member of the support team reads it and sends it.
Can it close or update a ticket?
No. It reads tickets and does not write to the ticket system.
What does it base a draft on?
Your help articles and past cases. Each draft lists the sources it used, so the person sending it can check them.
Where is customer data processed?
On your own infrastructure, with a model you serve. Tickets are not sent to a public AI service.